表征真实骨料形态的三维数值骨料建模

利用图像处理技术将骨料截面轮廓处理为二值图像,识别出各独立轮廓的形态特征后获取Hurst指数值的统计学规律,并使用改进的分形布朗运动(FBM)生成随机的数值骨料截面轮廓.使用该轮廓作为垂直于构建轴线的三维骨料上下两部分的连接面,基于改进的随机中点位移法,分别建立三维骨料的上曲面和下曲面并组装成完整的三维骨料模型.按相同纵长比生成的数值骨料与真实骨料具有相近的轮廓、扁平比和球度,且随机生成的数值骨料样本库与实测的真实骨料样本库具有类似的体积分布规律.按级配生成数值骨料并投放生成圆柱形沥青混凝土数值试件,相同工况下的仿真结果与单轴压缩蠕变试验数据的趋势吻合.

3D Numerical Modeling Methods to Characterize Real Aggregate Shapes

A statistical rule of the Hurst exponent was obtained based on the 2D morphology characteristic of independent aggregate sections, which were processed into binary images by image processing technologies. Meanwhile, a random contour of one aggregate section can be created by the improved fractional Brownian movement(FBM). This contour, which is perpendicular to the construction axis, is regarded as the connecting surface between the upper and lower surfaces of an aggregate waiting for construction. The upper and lower surfaces, which are respectively generated based on the improved random midpoint displacement method, are assembled into a complete numerical model of the 3D aggregate. Based on the same lengthwise ratio, the constructed numerical aggregates have the similar contours, flakiness ratios and sphericities with those of the real aggregates. The volume distribution law of random numerical aggregates has a resemblance to that of measured real aggregates. The constructed numerical aggregates based on aggregate gradations of a test were placed into a cylindrical asphalt concrete specimen. Under the same working condition, the simulation result is in agreement with data of the uniaxial compression creep test.